Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 224 (2018) 116–131

Formation of chondrules in a moderately high dust enricheddisk: Evidence from oxygen isotopes of chondrules from the

Kaba CV3 chondrite

Andreas T. Hertwig ⇑, Celine Defouilloy, Noriko T. Kita

WiscSIMS, Department of Geoscience, University of Wisconsin-Madison, Madison, WI 53706, USA

Received 6 April 2017; accepted in revised form 11 December 2017; available online 16 December 2017

Abstract

Oxygen three-isotope analysis by secondary ion mass spectrometry of chondrule olivine and pyroxene in combination withelectron microprobe analysis were carried out to investigate 24 FeO-poor (type I) and 2 FeO-rich (type II) chondrules fromthe Kaba (CV) chondrite. The Mg#’s of olivine and pyroxene in individual chondrules are uniform, which confirms that Kabais one of the least thermally metamorphosed CV3 chondrites. The majority of chondrules in Kaba contain olivine and pyrox-ene that show indistinguishable D17O values (= d17O � 0.52 � d18O) within analytical uncertainties, as revealed by multiplespot analyses of individual chondrules. One third of chondrules contain olivine relict grains that are either 16O-rich or 16O-poor relative to other indistinguishable olivine and/or pyroxene analyses in the same chondrules. Excluding those isotopicallyrecognized relicts, the mean oxygen isotope ratios (d18O, d17O, and D17O) of individual chondrules are calculated, which areinterpreted to represent those of the final chondrule melt. Most of these isotope ratios plot on or slightly below the primitivechondrule mineral (PCM) line on the oxygen three-isotope diagram, except for the pyroxene-rich type II chondrule that plotsabove the PCM and on the terrestrial fractionation line. The D17O values of type I chondrules range from ��8‰ to ��4‰;the pyroxene-rich type II chondrule yields �0‰, the olivine-rich type II chondrule ��2‰. In contrast to the ungrouped car-bonaceous chondrite Acfer 094, the Yamato 81020 CO3, and the Allende CV3 chondrite, type I chondrules in Kaba onlypossess D17O values below �3‰ and a pronounced bimodal distribution of D17O values, as evident for those other chondrites,was not observed for Kaba.

Investigation of the Mg#-D17O relationship revealed that D17O values tend to increase with decreasing Mg#’s, similar tothose observed for CR chondrites though data from Kaba cluster at the high Mg# (>98) and the low D17O end (�6‰ and�4‰). A mass balance model involving 16O-rich anhydrous dust (D17O = �8‰) and 16O-poor water ice (D17O = +2‰) in thechondrule precursors suggests that type I chondrules in Kaba would have formed in a moderately high dust enriched proto-planetary disk at relatively dry conditions (�50–100� dust enrichment compared to Solar abundance gas and less than 0.6�ice enhancement relative to CI chondritic dust). The olivine-rich type II chondrule probably formed in a disk with higher dustenrichment (�2000� Solar).� 2017 Elsevier Ltd. All rights reserved.

https://doi.org/10.1016/j.gca.2017.12.013

0016-7037/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: hertwig@wisc.edu (A.T. Hertwig).

1. INTRODUCTION

Chondrules – a main constituent of most chondriticmeteorites – are spherical, melted objects that formed bytransient high-temperature events in the proto-planetarydisk (e.g., Hewins, 1996; Rubin, 2000; Connolly and

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 117

Desch, 2004; Ciesla, 2005; Morris et al., 2012). Since therecognition of isotopic anomalies in the Allende CV chon-drite in the early 1970s (Clayton et al., 1973, 1977), numer-ous oxygen isotope studies on primitive carbonaceouschondrites have targeted the variability among bulk mete-orites, as well as those of CAIs and chondrules (e.g.,Clayton, 1993; Clayton and Mayeda, 1999; Jones et al.,2004; Krot et al., 2006; Yurimoto et al., 2008, and refer-ences therein). In general, oxygen isotope ratios of chon-drules in carbonaceous chondrites plot in the oxygenthree-isotope diagram below the terrestrial fractionation(TF) line and along a slope �1 line, for instance parallelto the CCAM (carbonaceous chondrite anhydrous miner-als) line, while those in ordinary chondrites cluster slightlyabove the TF line (e.g., Clayton, 1993).

By using secondary ion mass spectrometers (SIMS),high precision (sub-‰) mineral-scale isotope data of chon-drules have become increasingly available in recent years(e.g., Chaussidon et al., 2008; Kita et al., 2010; Connollyand Huss, 2010; Rudraswami et al., 2011; Ushikuboet al., 2012; Schrader et al., 2013, 2014, 2017; Tenneret al., 2013, 2015; Nagashima et al., 2015). In contrast tobulk analyses, in-situ measurements of primary chondruleminerals, such as olivine, low-Ca pyroxene and mesostasisphases allow for an evaluation of the isotopic homogeneityof single chondrules. A growing number of studies haveestablished that most chondrules are internally homoge-neous in terms of their D17O (= d17O – 0.52 � d18O) valuesexcept for minor occurrence of relict olivine grains (e.g.,Kita et al., 2010; Rudraswami et al., 2011; Tenner et al.,2013, 2015, 2017; Schrader et al., 2017). In particular,according to the analyses of ungrouped carbonaceouschondrite Acfer 094, in which both thermal metamorphismand aqueous alteration in the parent body was minimal,Ushikubo et al. (2012) clearly demonstrated that oxygenisotope ratios of olivine and pyroxene phenocrysts areindistinguishable from those of glassy mesostasis. Presum-ing that this is the general case, i.e., chondrule phenocrystsand mesostasis have indistinguishable D17O values at thetime of chondrule formation, it is now possible to deducethe composition of the isotopically homogenized melt bymeasuring chondrule phenocrysts in other chondrites evenif oxygen isotope ratios of glassy mesostasis were altereddue to parent body processes (e.g., in Semarkona LL3 byKita et al., 2010).

It has been suggested that chondrules formed in an opensystem with respect to major oxides, such as MgO andSiO2, which evaporated and re-condensed during melting(e.g., Tissandier et al., 2002; Alexander, 2004; Nagaharaet al., 2008). Under a dust-enriched environment (e.g.,Ebel and Grossman, 2000), the ambient gas present duringchondrule formation would have oxygen isotope ratios sim-ilar to those of average solid precursors, since oxygen in theambient gas would predominately originate from these pre-cursors (e.g., Ushikubo et al., 2012). Further, Ushikuboet al. (2012) suggested that the internally homogeneous oxy-gen isotope ratios within a single chondrule were the resultof isotope exchange between ambient gas and chondrulemelt that occurred rapidly due to evaporation and re-condensation of major oxides from the chondrule melt.

The Mg#’s of chondrules, defined as the molar MgO/(MgO + FeO) % of mafic chondrule minerals, mainlydepend on the redox conditions during chondrule-formation (Ebel and Grossman, 2000). Oxygen fugacitiesrequired to form FeO-poor (type I) or FeO-rich (type II)chondrules in carbonaceous chondrites (log fO2: up toiron-wustite buffer for type II; IW – 6 to �2 for type I chon-drules; e.g., Ebel and Grossman, 2000; Tenner et al., 2015)are higher than estimates for the Solar nebula (log fO2: �IW – 6 at 1600 K; e.g., Krot et al., 2000). The more oxidiz-ing redox conditions were likely imposed by enhancement(relative to Solar abundances) of dust particles and H2Oice (e.g., Ebel and Grossman, 2000; Fedkin andGrossman, 2006, 2016; Schrader et al., 2013; Tenneret al., 2015). In carbonaceous chondrites, many studieshave found type I chondrules to be generally 16O-rich com-pared to type II chondrules (e.g., Kunihiro et al., 2004,2005; Connolly and Huss, 2010; Schrader et al., 2013). Fur-ther, Ushikubo et al. (2012) and Tenner et al. (2013)observed a bimodal distribution of D17O values that isrelated to Mg#’s of chondrules in Acfer 094 and Y-81020(CO3), respectively. The majority of chondrules possessD17O values of ��5‰ and Mg#>97; other chondrulesshow D17O values of ��2‰ and a wide range of Mg#’s(100–40). In CB and CH chondrites, the majority of chon-drules are type I with D17O values of ��2‰ (Krot et al.,2010), while type II chondrules have D17O values of�+1.5‰ (Nakashima et al., 2010). In the case of CR chon-drites, the majority of chondrules are type I showing a lar-ger range of D17O values from �6‰ to �1‰, while minortype II chondrules span from �2‰ to 0‰ (e.g., Connollyand Huss, 2010; Schrader et al., 2013, 2014, 2017; Tenneret al., 2015).

The overall tendency of higher chondrule D17O valuesin combination with lower Mg#’s across carbonaceouschondrites suggests the existence of 16O-poor water iceamong chondrule precursors in carbonaceous chondriteforming regions (e.g., Connolly and Huss, 2010;Ushikubo et al., 2012; Schrader et al., 2013). By detailedexamination of Mg#’s and corresponding D17O valuesamong type I chondrules in CR chondrites, Tenner et al.(2015) observed a monotonic increase in D17O values withdecreasing Mg#, which was never observed in other chon-drite groups before. Tenner et al. (2015) uses an oxygenisotope mass balance model in combination with expres-sions that link dust enrichment and abundance of waterice to the fO2 of the chondrule melt in order to explainD17O values and Mg#’s of chondrules. By applyingD17O values of �6‰ and +5‰ for anhydrous dust andwater ice, respectively, the model estimated that type Ichondrules in CR formed under 100–200� dust-enrichments relative to solar composition gas with vari-able amounts of water ice from 0 to 0.8� the nominalwater ice content of CI chondritic dust. Tenner et al.(2015) stated that CR chondrites, Acfer 094, CO, and per-haps CV chondrites all contain chondrules with highMg#’s (>98) and D17O values ranging from �6‰ to�4‰. This range of compositions was also found to bethe dominant chondrule type in the Y-82094 ungroupedcarbonaceous chondrite (Tenner et al., 2017).

118 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

The literature data about the Mg#-D17O relationship forchondrules from CV chondrites, as summarized in Tenneret al. (2015), requires re-examination because of the follow-ing reasons. Earlier SIMS studies are mostly not at sub-‰precision level and mainly include olivine analyses and onlya limited amount of low-Ca pyroxene analyses (e.g., Choiet al., 2000; Jones et al., 2004; Libourel and Chaussidon,2011), hence, rendering it difficult to evaluate the mineral-scale isotopic homogeneity of these chondrules.Chaussidon et al. (2008) measured oxygen isotope ratiosof several chondrule constituents in Vigarano CV3 and,for a somewhat limited number of chondrules, in theMokoia and Efremovka CV3 chondrites; however, corre-sponding information about the mineral chemistry, espe-cially the Mg#’s for each individual chondrule, were notprovided. Conversely, a complete data set of oxygen iso-tope ratios of chondrule phases, including olivine, low-Capyroxene, and, where applicable, spinel and plagioclasewas published by Rudraswami et al. (2011) for Allende,but the evaluation of the Mg#-D17O relationship is madedifficult by significant thermal metamorphism experiencedby this chondrite (e.g., Krot et al., 1995; Bonal et al.,2006). Since diffusion of Mg and Fe in olivine and pyroxeneis considerably faster than diffusion of oxygen (Dohmenand Chakraborty, 2007; Farver, 2010), chondrules morelikely preserve primary oxygen isotopic compositions thanMg#’s when subjected to thermal metamorphism. Forexample, even though many olivine grains show higherFeO contents (lower Mg#’s) than coexisting low-Ca pyrox-enes in chondrules of the Allende CV3, most chondrules arestill homogeneous in oxygen isotope ratios at analyticalprecisions (Rudraswami et al., 2011).

Here we present SIMS oxygen three-isotope analysis ofchondrules from Kaba in order to understand the chon-drule formation environment and oxygen isotope reservoirsfor the CV chondrite-forming region. Kaba is one of theleast thermally metamorphosed CV3 chondrites (Kimuraand Ikeda, 1998; Krot et al., 1998; Grossman andBrearley, 2005; Bonal et al., 2006; Busemann et al., 2007)and a member of the Bali-like oxidized subgroup(McSween, 1977). In particular, representative dust-enrichment and ice-enhancement factors of the local diskregion are evaluated using the oxygen isotope mass balancemodel of Tenner et al. (2015) by applying model parameterssuitable for the CV chondrule-forming region. UnlikeAllende, chondrules in Kaba likely preserve primaryMg#’s in addition to D17O values that are representativefor conditions during chondrule formation, an assumptiontested by comparing Mg# of olivine and pyroxene withineach chondrule.

2. ANALYTICAL TECHNIQUES

2.1. Electron microscopy

One thin section of Kaba (USNM 1052-1) was imaged(BSE, SE) and the petrography of chondrules investigatedwith a Hitachi S-3400N scanning electron microscope(SEM). The chemical composition of olivine and pyroxenegrains were determined by a Cameca SX-51 electron micro-

probe (20 nA, 15 keV, fully focused beam); plagioclase andhigh-Ca pyroxene of the mesostasis were analyzed by aCameca SXFive (10 nA, 15 keV, 10 lm beam). The SEMand both electron microprobes are hosted at the Depart-ment of Geoscience at UW-Madison. On both electronmicroprobes, concentrations of the oxides SiO2, TiO2,Al2O3, Cr2O3, MgO, FeO, MnO, CaO, K2O, and Na2Owere acquired and counting times on the peak and back-ground were set to 10 s and 5 s, respectively. The 3r detec-tion limits (wt%) for these oxides (in the order mentionedabove) were at most: 0.05, 0.06, 0.04, 0.07, 0.05, 0.07,0.07, 0.04, 0.03, and 0.06, respectively. Data reduction,including ZAF/u(qz) corrections, was done with the‘‘Probe for EPMA” software suite (Donovan, 2015). Thefollowing standards were used for olivine and pyroxeneanalyses (on Cameca SX-51): Ti: rutile, Al: jadeite, Cr: syn-thetic Cr2O3, Fe: synthetic hematite, Mn: manganese oli-vine, Ca: wollastonite, K: microcline, Na: jadeite, anddepending on mineral type and composition: Si: syntheticenstatite, synthetic forsterite, Fo83, manganese olivine;Mg: synthetic enstatite, synthetic forsterite, Fo83, amphi-bole. Analyses of plagioclase and high-Ca pyroxene in thechondrule mesostasis were performed on a Cameca SXFive,using the same suite of standards except for the followingelements: Si: synthetic enstatite, An78, An67; Al: jadeite,An95; Fe: fayalite; Ca: An78, An95; Na: jadeite, albite, An78.

2.2. SIMS oxygen isotope analyses

The in-situ oxygen three-isotope analysis of olivine andpyroxene was carried out during two separate sessions withthe Cameca IMS 1280 SIMS at the UW-Madison usingmulti-collector Faraday cups and following analytical pro-tocols similar to those of Kita et al. (2010) and Tenner et al.(2013). In both sessions, Cs+ primary ion intensity was setto �3 nA in order to generate a secondary 16O� intensity of�3 � 109 cps (counts per second). During the first session,the primary beam was tuned to form an ellipsoid-shapedspot of 14 � 10 lm, similar to the condition used in previ-ous studies (e.g., Tenner et al., 2015, 2017). In the secondsession, the primary beam aperture was enlarged (>200mm, normally 100–150 mm) and resulted in a diffused beamshape (Gaussian beam in both sessions). Consequently, wereduced the beam size to �10 mm but rastered 5 � 5 lmover the sample surface to produce a roundish spot of 12lm diameter. The intensity of 16O1H� was monitored auto-matically at the end of each analysis by using a X-deflectorbetween sector magnet and detectors in order to correct thecontribution from tailing of the 16O1H� peak to the 17O�

signal. The correction was always insignificant (�0.1‰).Typically, 12–18 analyses of unknowns were bracketed

by 8 analyses (4 before and 4 after unknowns) of the SanCarlos olivine standard (d18O = 5.32‰ VSMOW, Kitaet al., 2010) to monitor the drift of instrumental bias andthe external (spot-to-spot) reproducibility. External repro-ducibility, which represents the uncertainty of individualanalyses of unknowns (Kita et al., 2009), was typically�0.4‰, �0.3‰, and �0.4‰ (2SD) for d17O, d18O, andD17O, respectively. Olivine and pyroxene form solid solu-tions series that show systematic difference in instrumental

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 119

biases in d18O (e.g., Tenner et al., 2013), which are cali-brated using three olivine (Fo100, 89, 60) and three low-Capyroxenes (En97, 85, 70) and one diopside (En50Wo50) stan-dard (Kita et al., 2010; Nakashima et al., 2013) that coverthe range of chemical compositions observed in olivineand pyroxene in Kaba chondrules.

2.3. Selection of chondrules and positions of SIMS analysis

To minimize the sampling bias that is potentially intro-duced by preferential selection of chondrules by e.g., shape,texture, or degree of alteration, every chondrule larger thanor equal to �750 lm in diameter (or longest direction) wasselected from a single polished thin section of Kaba(USNM 1052-1) for oxygen isotope analysis based on theevaluation of high-resolution BSE and SE images of theentire thin section. Due to scarcity of type II chondrulesin Kaba, one small (�150 mm) chondrule comprisingFeO-rich olivine was included for SIMS analysis as wellas one smaller chondrule (�600 lm) comprising FeO-richpyroxene. All selected chondrules are labeled (K1 to K26)in the BSE mosaic image of the Kaba USNM 1052-1 thinsection shown in Appendix EA1. At this point, exploratorySEM-EDS measurements were carried out to examinechemical zoning of minerals and to identify opaque phases.Per chondrule, at least 8 locations that are larger than 15mm and free of cracks and inclusions were selected forSIMS analyses, preferably 4 for olivine and 4 for pyroxene,to be able to evaluate the isotopic homogeneity of individ-ual chondrules. High-Ca pyroxene or plagioclase in themesostasis of chondrules were not analyzed for oxygenthree-isotopes, because they are usually smaller than 15mm. The chemical compositions of minerals at each positionwere determined by EPMA and utilized later for instrumen-tal bias correction of SIMS analyses. Finally, after SIMSanalysis, each individual SIMS pit was imaged by SEM(see Appendix EA2) to evaluate whether cracks or inclu-sions were hit and whether the desired position was accu-rately sampled by the ion beam. Individual SIMSanalyses were rejected from results if the primary beamoverlapped two different mineral grains or an elevatedOH signal in combination with visible cracks or cavitieswithin pits indicate a compromised analysis.

2.4. Data reduction for host D17O values of individual

chondrules

In order to estimate representative oxygen isotope ratiosof the last chondrule melt, Ushikubo et al. (2012) andTenner et al. (2013) established a data reduction schemeto calculate the host D17O value of a single chondrule asthe mean value of multiple olivine and pyroxene analysesthat are within a critical value from the mean. The samedata reduction scheme was applied for chondrules in Kababy using 0.6‰ (D17O) as the critical value, which is themean 3SD of the bracketing San Carlos olivine standardduring the 2 SIMS sessions. In a first step, the mean D17Ovalue from multiple analyses of an individual chondrulewas calculated. If all data are within ±0.6‰ from the mean,this mean D17O value is considered to represents that of the

host chondrule. The host D17O value should be determinedby at least 2 data points. In a second step, for chondrulesthat contain data exceeding ±0.6‰ from the mean, a subsetof data is selected, which preferably includes all pyroxeneanalyses and a new mean is calculated, only including thisnew subset. Subsequently, the subset is tested again forthe variability criterion. The excluded olivine analyses,deviating more than 0.6‰ from the mean, are potentialrelict grains that might not reflect D17O values of the finalchondrule melt.

The host d17O and d18O values are calculated with thesame subset of analyses that were used for calculating thehost D17O values. Uncertainties (at 95% confidence level)reported for host d17O and d18O values comprise the prop-agation of three types of uncertainties related to the analy-sis of unknowns, the instrumental bias correction based onthe bracket standard, and the ultimate uncertainty of SIMSoxygen isotope analyses: Uncertainties (unc.) are calculatedas the sum of those three components that are (i) twice the

standard error 2 SDffiffin

p� �

of those analyses that constitute the

mean, where SD is standard deviation of the unknownanalyses or that of the standard bracket whichever larger,(ii) twice the standard error of the analyses of the corre-sponding San Carlos olivine bracket, and (iii) a fixed valueto account for possible mass-dependent fractionation inmean isotope ratios because of sample geometry and topog-raphy, as well as the reproducibility of calibration stan-dards (±0.3‰ for d18O, ±0.15‰ for d17O, Kita et al.,2009, 2010). Uncertainties of host D17O values only con-sider components (i) and (ii).

3. RESULTS

3.1. Petrography of chondrules

In the Kaba thin section, 7 type IA (<10% modal abun-dance of low-Ca pyroxenes), 3 type IB (<10% olivine), 14type IAB, 1 type IIA, and 1 type IIB were analyzed for oxy-gen isotope compositions. The two individual chondrulesK8 and K9 are part of one compound object and discussedtogether as chondrule K8 + 9 (Fig. 1a). Petrographicdescriptions of each chondrule analyzed in this study is pre-sented in the Appendix EA3. Chondrule sizes range from0.6 to 3 mm (diameter or longest direction); the type IIAchondrule is considerably smaller (0.15 mm). Throughoutthis paper no distinction is made between complete chon-drules and chondrule fragments. General textures of mostinvestigated chondrules are best described as porphyritic.

Some type IAB chondrules show a predominately con-centric layering. In particular, concentric layering withinindividual chondrules is well developed in the compoundchondrule K8 + 9 (Fig. 1a), as well as in chondrules K1(Fig. 1b) and K20 (Appendix EA1, Page 20). In these exam-ples, inner portions of chondrules comprise olivine,mesostasis, and in the case of the individual chondruleK8 also abundant opaque phases. Inner portions are man-tled by pyroxenes, often containing olivine inclusions, andare enclosed with fine-grained rims dominated by pyroxeneand dispersed opaque phases. The texture and mineralogy

Fig. 1. Representative BSE images of chondrules from the Kaba USNM 1052-1 thin section showing SIMS pits for oxygen three-isotopeanalyses (except for K1). (a) Compound chondrule K8 + 9 (IAB) enclosed in igneous rim comprising pyroxene, olivine and opaque phases. (b)K1 (IAB) composed of coarse-grained olivine and pyroxene phenocrysts, as well as anorthite and high-Ca pyroxene in mesostasis at thechondrule center; enclosed in relatively fine-grained and opaque phases-rich rim. (c) K22 (IAB) contains coarse-grained olivine and low-Capyroxene. Altered mesostasis comprises phyllosilicates and high-Ca pyroxenes, latter forming euhedral overgrowths on low-Ca pyroxenes. (d)Low-Ca pyroxene poikilically encloses olivine in K11 (IB). (e) K16 (IA) comprises small amount of mesostasis and intermediate pyroxenealongside olivine. (f) FeO-rich low-Ca pyroxenes (En85) in K25 (IIB) possess overgrowth of FeO-rich olivine (Fo83). Ol: olivine, Lpx: low-Capyroxene, HPx: high-Ca pyroxene, IntPx: intermediate pyroxene, An: anorthite, Phy: phyllosilicates; scale in all images: 500 lm.

120 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 121

of the irregular-shaped chondrule K12 (see Appendix EA1,Page 12) resembles those of these fine-grained rims. Tex-tures of the type IAB chondrules K3 and K19 (see Appen-dix EA1) indicate a similar concentric layering but fine-grained rims are absent.

In general, olivine occurs in type I chondrules (i) asanhedral grains (e.g., K3, K5, K8 + 9) or, rarely, as euhe-dral phenocrysts (K1) and small euhedral crystals inmesostasis (K4), (ii) in the form of irregular-shaped olivinemasses (e.g., K19; K22, Fig. 1c), and (iii) as small euhedralgrains poikilically enclosed in low-Ca pyroxene (e.g., K1,K24; K11, Fig. 1d). In type IA chondrules, olivine grainscan be coarse-grained and embedded in varying amountsof interstitial mesostasis (K7, K10; K16, Fig. 1e). Chon-drule K6 (IA) comprises olivine that contains metal grains,typical for ‘‘dusty olivine” relicts. Except for its occurrencein fine-grained rims, low-Ca pyroxene is usually subhedral(or even euhedral) and often exhibits a poikilitic texturecaused by small individual inclusions (e.g., K11, K15,K24) or clusters of olivine grains (K13). The type IIB chon-drule K25 comprises subhedral pyroxene that exhibits thin(<20 lm) overgrows of olivine (K25, Fig. 1f).

Chondrules contain varying amounts of opaque phases(e.g., high abundance: K11, K12, K14, K23; low: K3, K7,K16, K22) that are evenly distributed within chondrules(e.g., K11, K12) or localized at their margins (e.g., K8 +9, Fig. 1a; K16, Fig. 1e). In chondrule K8 + 9, larger blobsof opaque phases are abundant at the interface of chon-drule interior and rim (Fig. 1a). As confirmed by SEM-EDS analyses, opaque phases are mainly magnetite, Fe-Ni sulfides, and, rarely, Fe-Ni alloys (only in K1). Themesostasis is altered in most of the chondrules and com-prises abundant phyllosilicates that probably replacedglassy mesostasis and plagioclase (Fig. 1c). High-Ca pyrox-ene often forms euhedral overgrowths on low-Ca pyroxenes(Fig. 1c), sometimes also worm-like intergrowth textureswith plagioclase (see Appendix EA1, Page 14).

Fig. 2. Oxygen three-isotope diagram showing olivine (n = 124)and pyroxene (n = 95) analyses from chondrules in Kaba. Mostanalyses plot on the primitive chondrule mineral (PCM; Ushikuboet al., 2012) line or in between the PCM and CCAM (Clayton et al.,1977) lines; analyses of chondrule K25 plot above PCM and onterrestrial fractionation (TF) line. YR: Young & Russel line(Young and Russell, 1998).

3.2. Mineral chemistry of olivine, pyroxene, and mesostasis

phases

Mg#’s of olivine in type I chondrules are restricted to anarrow range between 98.5 and 99.8 (99.3 ± 0.58, 2SD) andzoning caused by thermal metamorphism is insignificant(complete set of EPMA analyses in Appendix EA4). ThinFeO-rich halos around opaques or small holes in olivine,once filled by opaque minerals, are evidence for small-scale diffusion and excluded from analyses. Some olivinein chondrule K1 shows higher than typical Cr2O3 (up to0.69 wt%) and MnO (up to 0.49 wt%) values compared tothe mean of all olivine analyses (Cr2O3: 0.36 ± 0.27 wt%;MnO: 0.18 ± 0.25 wt%; 2SD). Chondrules K3, K7, andK22 contain olivine grains that yield CaO contents slightlyhigher than 0.5 wt% (up to 0.59 wt%). Olivine in the typeIIA chondrule (K26) is zoned in Fe content and yieldsMg#’s of 66.2 and 57.7 in the center and close to margins,respectively. The thin olivine overgrowths on pyroxene inthe type IIB chondrule (K25) possess Mg#’s of about 83.

Low-Ca pyroxene (Wo<3) is Fe-poor (mean Mg# type I:99.1 ± 0.51, 2SD) and shows the characteristic twinning of

clinoenstatite when observed under crossed polarizers. Inaddition to low-Ca pyroxene, some type I chondrules(K5, K6, K8 + 9, K10, K13, K16) contain ‘‘intermediatepyroxene” (IntPx) often associated with mesostasis and dis-tinguished from low-Ca pyroxene by higher Wo contents(Wo3-5); lamellar twins are often absent in intermediatepyroxene. The fine-grained rim enclosing chondrules K8and K9 comprises low-Ca and intermediate pyroxene. Inchondrule K10, intermediate and high-Ca pyroxenes inthe mesostasis are the only pyroxenes; in chondrule K16,low-Ca pyroxene is only present in insignificant amountsat chondrule margins whereas intermediate pyroxene fillsinterstitials of coarse olivine grains (Fig. 1e). Pyroxene inthe type IIB chondrule (K25, Fig. 1f) is En85Fs14Wo1(Mg#: 85.7). In general, high-Ca pyroxenes (En55-65Fs<2-Wo

34-44) that coexist with plagioclase (An91-98) in the

mesostasis of several chondrules can be considered as alu-minian augites (Al > 0.1 apfu, atoms per formula unit). Inthese aluminian augites, the Ca(R3+)AlSiO6 component,where R3+ is predominately Al, can exceed 15 mol% in afew cases. Ti contents of high-Ca pyroxenes are low(<0.05 apfu).

3.3. Oxygen three-isotope ratios

A total of 235 SIMS oxygen isotope analyses of olivine(133) and pyroxene (102) were performed. Thereof 16 hadto be rejected from final results, mostly because (i) the spotarea covered two different phases (e.g., K1: 98.Ol, 104.Lpx;see Appendix EA2, Page 1), (ii) abundant cracks (e.g., K4:234.Lpx), (iii) the beam hit cavities in grains (e.g., K23: 309.Ol) or (iv) due to other surface imperfections. SIMS

122 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

analyses are referred to in this paper by citing the chondruleidentifier, followed by the analysis number and an abbrevi-ation of the mineral analyzed. The complete oxygen three-isotope dataset of this study, including rejected analyses, isprovided in Appendix EA5. Individual oxygen three-isotope diagrams of all analyzed chondrules are given inAppendix EA6.

Most oxygen isotope ratios of olivine and pyroxene ploteither on the PCM or in between the PCM and CCAM lines(Fig. 2). The d18O and d17O values of all individual analysesrange from ��36‰ to +4‰ and �38‰ to +2‰, respec-tively (D17O � �19‰ � 0‰). Some of the most 16O-richanalyses (e.g., D17O � �15‰, K4: 235.Ol, 236.Ol) showinternal errors (derived from cycle-by-cycle variationswithin a single spot analysis) that are larger than typical(<0.5‰), indicating heterogeneous oxygen three isotoperatios within an analyzed volume. Typical olivine andpyroxene in type I chondrules yield d18O and d17O valuesin between �12‰ and �1‰ and from �15‰ to �5‰,respectively (D17O � �8‰ to �3‰). Pyroxene in the typeIIB (d18O: +3.7 � +4.0‰, d17O: +2.0 � +2.1‰, D17O:0.0 � +0.2‰) and olivine in the FeO-rich type IIA chon-drule (d18O: +2.6 � +2.9‰, d17O: �1.1 to �0.4‰, D17O:�2.5‰ to �1.9‰) are 16O-poor compared to those in typeI chondrules. Analyses of olivine overgrowths on pyroxeneof the type IIB chondrule (K25: 327.Ol, 328.Ol) have beenrejected because of irregular SIMS pits, yet their D17O val-ues (D17O: 0.09 ± 0.01‰, n = 2, 2SD) are indistinguishablewithin analytical uncertainty from those of the pyroxene(0.06 ± 0.19‰, n = 3, 2SD) in the same chondrule.

3.4. The presence of isotopic relict grains and host chondrule

D17O values

Fig. 3 provides oxygen three-isotope diagrams forselected chondrules and Fig. 4 shows D17O values of olivineand pyroxene as well as mean chondrule D17O values. Forall chondrules examined, at least 3, but usually more than7 analyses are indistinguishable within analytical uncertain-ties (Fig. 3a–g). 11 out of 25 chondrules comprise olivineanalyses that are variable in D17O values beyond the thresh-old, and some of them are considered to represent analysesof relict grains. Most relict olivine grains are 16O-rich rela-tive to their host, but in some chondrules (see Fig. 4; K18,K16) relict olivine grains are 16O-poor relative to the corre-sponding host. Olivine analyses in chondrules K5, K13, andK16 form clusters with mean D17O values that are distin-guishable from those of the corresponding pyroxene clus-ters. Therefore, in these cases, the host D17O values aredetermined by using only pyroxene analyses although apart of the olivine analyses in chondrules K5 and K13 arewithin the variability threshold. In chondrules K6, K8 +9, and K24, the internal variability of D17O is marginalcompared to the threshold of 0.6‰.

Most chondrules in the Kaba 1052-1 thin section (21 outof 25) yield almost continuous host D17O values in between�6.1‰ and �3.9‰ (d18O: �6.9‰ to �2.1‰, d17O: �9.7‰to �5.0‰, see Fig. 4 and Table 1). Chondrules K18 (IA)and K23 (IAB) possess the lowest host D17O values of�8.3‰ (d18O: �12.0‰, d17O: �14.5‰) and �7.8‰ (d18O:

�7.0‰, d17O: �11.4‰), respectively. The type IIB chon-drule (K25) and the type IIA chondrule (K26) show themost 16O-poor compositions with host D17O values of0.1‰ (d18O: 3.8‰, d17O: 2.0‰) and �2.2‰ (d18O: 2.7‰,d17O: �0.8‰), respectively.

4. DISCUSSION

4.1. Evaluation of host chondrule oxygen isotope ratios

The primary aim of this SIMS oxygen isotope study ofolivine and low-Ca pyroxene phenocrysts is to determinethe oxygen isotope ratios of the last chondrule-formingmelt that would record the average oxygen isotope ratiosof the local dust-enriched disk. For clarity, the term ‘‘lastchondrule-forming melt” is used here to refer to the pro-duct of the last major melting event that is recorded in mostchondrules of one chondrite. These melts interacted withthe ambient gas before or during crystallization of chon-drule minerals (e.g., Tissandier et al., 2002; Hewins andZanda, 2012; Nagahara and Ozawa, 2012; Di Rocco andPack, 2015; Marrocchi and Chaussidon, 2015) and thereis evidence that chondrule phenocrysts and evolved meltswere not in chemical equilibrium (e.g., Libourel et al.,2006). This disequilibrium, in turn, raises some doubtswhether D17O values of olivine and low-Ca are actual rep-resentative for those of the last chondrule-forming melt.However, Ushikubo et al. (2012) showed for Acfer 094(ungr. CC), the most pristine carbonaceous chondriteknown to date, that chondrule phenocrysts (excludingrelicts) and mesostasis phases such as high-Ca pyroxenes,plagioclase, and glass are indistinguishable in respect toD17O values. Also, high-Ca pyroxenes, plagioclase, and oli-vine show indistinguishable D17O values in chondrules fromtwo CR chondrites (Tenner et al., 2015) and Yamato-82094(ungr. CC, Tenner et al., 2017). Moreover, it is likely thatsystematically higher D17O values of chondrule glass inSemarkona (LL, Kita et al., 2010) or plagioclase in Kaba(CV, Krot and Nagashima, 2016) are due to parent bodyprocesses and not caused by gas-melt exchange during pla-gioclase crystallization or glass formation.

Porphyritic chondrules can contain mineral grains thatdidn’t completely melt during the chondrule-forming eventand are identified as relict grains either by chemical compo-sitions and/or isotope ratios (e.g., Jones et al., 2004;Kunihiro et al., 2004; Krot et al., 2006; Berlin et al.,2011; Rudraswami et al., 2011; Ushikubo et al., 2012;Schrader et al., 2013; Tenner et al., 2013). Relict olivinegrains predate ‘‘host” minerals that crystallized from thefinal chondrule melt (Nagahara, 1981; Jones, 1996;Wasson and Rubin, 2003). Classical examples of those relictolivine are 16O-rich forsteritic cores in type II chondrules ofcarbonaceous chondrites (e.g., Yurimoto and Wasson,2002; Kunihiro et al., 2004, 2005; Rudraswami et al.,2011; Ushikubo et al., 2012; Schrader et al., 2013; Tenneret al., 2013). However, in most of the cases, isotopically dis-tinct olivine relict grains are chemically and petrographi-cally indistinguishable from other olivine of the samechondrule and only identified by SIMS analyses at sub-‰precisions (e.g., Rudraswami et al., 2011; Ushikubo et al.,

Fig. 3. Oxygen three-isotope diagrams of individual olivine and pyroxene analyses of representative chondrules in Kaba. Error bars showexternal reproducibility (2SD) obtained by analyses of the bracketing olivine standard. (a–g) Individual analyses are indistinguishable in termsof their D17O values within the variability threshold of ±0.6‰. (h) One olivine analysis in chondrule K14 slightly exceeds variability threshold.(i and j) Olivine and pyroxene analyses form distinct clusters, while (relict) olivine are either 16O-rich (K13) or 16O-poor (K16) relative topyroxene. (k and l) Analyses of K6 and K8 + 9 show clearly resolvable variability in d18O and d17O values (2SD: 1.0–1.4‰) but D17O valuesare within variability threshold (±0.6‰). Both chondrules are marginally heterogeneous. In K6 (k), analyses located in right hand part ofchondrule are relatively more 16O-rich (circled data points). In K8 + 9 (l) pyroxene analyses in the rim and a few locations of the chondruleinterior (circled data points) are 16O-poor relative to the rest of the analyses.

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 123

Fig. 4. D17O values of olivine and pyroxene analyses per individual chondrule. Error bars show external reproducibility (2SD) obtained byanalyses of the bracketing standard. Chondrules sorted according to increasing host chondrule D17O values (stars) which are obtained byaveraging D17O values of isotopically homogeneous olivine and/or pyroxene analyses after excluding isotopic relicts (triangles). D17O values ofrelict olivine exceed the variability threshold (grey line; ±0.6‰, mean external precision on bracketing standard, 3SD). Most type I chondrulesshow host chondrule D17O values in between �6‰ and �4‰ as illustrated by the histogram on the right-hand side of the graph.

124 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

2012; Tenner et al., 2013, 2015, 2017). In order to deducerepresentative oxygen isotope ratios of the chondrule-forming environment, it has proven useful to define mean‘‘host” oxygen isotope ratios. These host values have anal-yses of those isotopic relict grains excluded from chondrulemeans (e.g., Kunihiro et al., 2004; Ushikubo et al., 2012;Tenner et al., 2017).

In 11 of 22 chondrules, analyzed olivine grains are indis-tinguishable from pyroxene in the same chondrule, i.e.,chondrules are relict grain-free and contain both minerals.The total number of chondrules cited here excludes chon-drules K23 and K25 where no valid pyroxene or olivineanalyses, respectively, are available as well as K26 that doesnot contain pyroxene. Moreover, 19 out of 22 chondrules,shown in Fig. 5a, are either relict grain-free or contain atleast one olivine with D17O values indistinguishable frompyroxene of the same chondrule; the remainder comprisesthree chondrules, K5, K13, and K16, where all olivine areconsidered to represent relict grains. Since olivine is the liq-uidus phase succeeded by low-Ca pyroxene during crystal-lization of type I chondrule melts, there is a possibilitythat a changing melt D17O value during crystallizationwould be reflected in differing D17O values of both minerals,e.g., higher D17O values of pyroxenes relative to those ofolivine (e.g., Chaussidon et al., 2008); such systematic rela-tionship was clearly not observed in this study. It is there-fore suggested that olivine and pyroxene crystallized froma homogenized melt in respect to D17O and that the meanchondrule D17O values of chondrule olivine and pyroxenecan be used as a proxy for those of the last chondrule-forming melt.

There could be, however, a tendency of olivine to showslightly higher d18O values than pyroxene of the same chon-drule, but caution is advised because variations are withinanalytical uncertainties (see Fig. 5b). Similar observationswere made earlier for different chondrites (Tenner et al.,2013, 2015) and attributed to evaporative loss of light iso-topes before olivine crystallization followed by re-condensation of light isotopes before or during pyroxenecrystallization (Tenner et al., 2015). Since such processesresult only in mass-dependent fractionation, chondruleD17O values and conclusions based thereof are not affectedby this type of gas–melt interaction.

Mean chondrule D17O values determined for Kaba areindependent of the modal abundance of olivine and pyrox-ene as shown in Fig. 6 which is in line with previous SIMSoxygen isotope studies on unequilibrated carbonaceouschondrites (Rudraswami et al., 2011; Ushikubo et al.,2012; Tenner et al., 2013, 2015). Apart from the Mg#’s, thisstudy found no connection between oxygen isotope ratiosand chemical compositions of olivine or pyroxene. Forexample, there exists no systematic difference of pyroxened17O, d18O, or D17O depending on the Wo content (i.e.,low-Ca pyroxene vs. intermediate pyroxene). In agreementwith previous SIMS oxygen isotope studies on chondrule incarbonaceous chondrites (e.g., Rudraswami et al., 2011;Ushikubo et al., 2012; Tenner et al., 2015), most host chon-drule D17O values plot on the PCM line or between thePCM and CCAM lines (see Fig. 6); notable exceptionsare chondrules K25 (type IIB) and K23 (IAB). In the lattercase (K23), isotopic compositions of olivine are potentiallyfractionated.

Table 1Average mineral chemistry and host oxygen isotope ratios (‰, VSMOW) of individual chondrules in the polished thin section of Kaba CV3(USNM 1052-1).

Chondrule Type Ol-ra Ol-ha Pxa Mg#b Woc d18Od d17Od D17Od

K1 (core) IAB – 12 9 99.0 1 �1.79 ±0.33 �5.15 ±0.24 �4.22 ±0.17K1 (rim) – – – 2 98.4 1 �1.79 ±1.36 �4.91 ±0.72 �3.98 ±0.29K1 (all) IAB – 12 11 98.8 1 �1.79 ±0.34 �5.13 ±0.24 �4.20 ±0.17K2 IA 2 6 2 99.2 2 (1–2) �3.07 ±0.50 �6.05 ±0.32 �4.45 ±0.29K3 IAB – 4 4 99.3 1 �2.12 ±0.42 �5.00 ±0.24 �3.89 ±0.26K4 IA 6 1 2 99.5 1 �6.91 ±0.49 �9.66 ±0.44 �6.07 ±0.28K5 IAB 3 – 5 99.3 3 (1–5) �4.46 ±0.47 �7.87 ±0.45 �5.56 ±0.28K6 IA – 6 4 99.3 3 (1–3) �2.45 ±0.44 �5.66 ±0.37 �4.39 ±0.21K7 IA – 5 3 99.4 1 �5.45 ±0.40 �8.11 ±0.22 �5.28 ±0.16K8 IAB – 6 4 99.3 1 �5.75 ±0.40 �8.78 ±0.31 �5.79 ±0.19K9 IAB – 2 3 99.4 2 (1–5) �5.34 ±0.86 �8.30 ±0.65 �5.53 ±0.27K8 + 9 (rim) – – – 4 99.1 4 (1–5) �4.47 ±0.36 �7.61 ±0.30 �5.28 ±0.23K8 + 9 (all) IAB – 8 11 99.3 2 (1–5) �5.37 ±0.46 �8.41 ±0.36 �5.62 ±0.19K10 IA 1 4 5 99.0 4 (4–5) �2.67 ±0.37 �5.77 ±0.29 �4.38 ±0.19K11 IB – 3 4 99.2 1 (1–3) –3.16 ±0.39 �6.33 ±0.26 �4.68 ±0.19K12 IAB 1 3 5 99.1 1 (1–2) �4.73 ±0.37 �7.85 ±0.26 �5.39 ±0.18K13 IB 3 – 5 99.0 2 (1–4) �3.08 ±0.49 �6.09 ±0.29 �4.48 ±0.23K14 IAB 1 3 4 99.1 1 �4.31 ±0.39 �7.64 ±0.29 �5.40 ±0.24K15 IB – 4 4 99.4 1 �5.85 ±0.41 �8.87 ±0.25 �5.83 ±0.18K16 IA 6 – 4 99.2 5 �4.62 ±0.42 �7.43 ±0.29 �5.03 ±0.27K17 IAB – 5 3 99.4 1 �4.90 ±0.36 �7.94 ±0.26 �5.40 ±0.21K18 IA 3 3 1 99.6 1 �11.96 ±0.41 �14.48 ±0.28 �8.26 ±0.14K19 IAB – 5 2 99.3 1 �4.37 ±0.65 �7.56 ±0.42 �5.29 ±0.18K20 IAB 2 3 2 99.2 1 �4.10 ±0.75 �7.41 ±0.61 �5.28 ±0.26K21 IAB – 5 3 99.2 1 �3.41 ±0.34 �6.97 ±0.30 �5.20 ±0.26K22 IAB – 4 4 98.8 1 �2.74 ±0.53 �5.79 ±0.38 �4.37 ±0.26K23 IAB – 4 – 99.5 – �7.02 ±0.31 �11.44 ±0.40 �7.79 ±0.37K24 IAB 1 3 4 99.4 1 �5.29 ±0.69 �8.48 ±0.63 �5.73 ±0.36K25 IIB – – 3 85.7 1 +3.82 ±0.35 +2.04 ±0.30 +0.06 ±0.24K26 IIA – 4 – 62.9e – +2.74 ±0.34 �0.80 ±0.38 �2.22 ±0.28

a Ol-r, Ol-h, and Px represent the number of SIMS analyses from relict olivine, host olivine, and low-Ca pyroxene, respectively.b Mean Mg# (MgO/(MgO + FeO) mol%) of olivine and pyroxene excluding relict olivine.c Mean Wo component of low-Ca pyroxenes in mol%; ranges shown in parenthesis, if significantly variable.d Quoted uncertainties are at 95% confidence level. Analyses of relict olivine are not included in the mean values.e Olivine zoned in Mg# (core � 66, rim � 58).

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 125

4.2. Consistent Mg#’s of chondrule olivine and pyroxene

Pristine oxygen isotope ratios and Mg#’s are a precon-dition for reliable conclusions about the chondrule-forming environment and consistent Mg#’s of olivine andpyroxenes are an important indicator whether thermalmetamorphism could have disturbed primary values.According to the equilibrium condensation model of Ebeland Grossman (2000), olivine and orthopyroxene initiallyshow similar Mg#’s of about 98 at the time the last meltdisappears (Ebel and Grossman, 2000, therein Fig. 8, dustenrichment factor: 100). Due to diffusive exchange withFe-rich minerals, olivine and pyroxene could become moreFeO-rich during parent body metamorphism. Importantly,the interdiffusion coefficient of Fe and Mg in olivine isabout 2 log units larger than in orthopyroxene (e.g.,Dohmen and Chakraborty, 2007; Dohmen et al., 2016,and references therein), i.e., diffusion rates are higher in oli-vine than in orthopyroxene. Therefore, low degrees of ther-mal metamorphism manifests itself first in elevated olivineMg#’s whereas pyroxene Mg#’s remain unchanged. Fur-ther, because diffusion of Fe and Mg is fast relative to that

of oxygen (Dohmen and Chakraborty, 2007), matchingMg# indicate that both minerals probably record primaryoxygen isotope ratios. However, it should be noted that flu-ids could have enhanced oxygen diffusion in those silicates(Farver, 2010, and references therein). In this study wefound consistent Mg#’s of most olivine pyroxenes withinindividual chondrules from Kaba, providing evidence forunaltered Mg#’s and D17O values. Those values can nowbe used to infer possible dust enrichment and ice enhance-ment factors of the CV chondrule-forming region.

4.3. Relating oxygen isotope compositions of chondrules to

levels of dust enrichment in the chondrule forming region

Tenner et al. (2015) proposed a simple yet powerfulmodel to relate the D17O values and Mg#’s of host chon-drules to the extent of dust enrichment and the contributionof distinct precursor oxygen reservoirs. The oxygen isotopecharacteristics of host chondrules are described by an oxy-gen isotope mass balance that involves Solar gas (D17O =�28.4‰), anhydrous silicate dust (�5.9‰) as well as organ-ics (+11.8‰) and H2O ice (+5.1‰) in the dust as precursor

Fig. 5. Mean olivine and pyroxene (a) D17O and (b) d18O values ofindividual chondrules. (a) Chondrule means (D17O) of olivine andpyroxene (excluding relict olivine) are shown for chondrules thathave indistinguishable olivine and pyroxene analyses. Error barsshow the propagated uncertainties of the mean values (seeSection 2.4. and Table 1). (b) Mean olivine d18O values are slightlyhigher than those of pyroxene but variations are within propagateduncertainties.

Fig. 6. Host chondrule isotope ratios (d18O, d17O, and D17O) andtextures of chondrules. Error bars represent propagated uncer-tainty of the mean values. (a) Oxygen 3-isotope diagram showingchondrule means. Reference lines are the same as those in Fig. 2.Host values of chondrules plot on the PCM line or slightly below.Exceptions are chondrules K23 (IAB) and K25 (IIB). (b) HostD17O values are independent of chondrule texture among type Ichondrules. Type II chondrules are 16O-poor compared to type Ichondrules.

126 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

oxygen reservoirs. At a given amount of water ice, chon-drule D17O values strongly increase with rising dust enrich-ment but continue to increase only marginally at dustenrichment factors higher than 100 because at these condi-tions, mass balance is dominated by nearly constantamounts of oxygen from the anhydrous silicate precursordust and the water ice.

For instance, if a chondrule formed in a region with 100times (CI) dust enrichment (H/O � 37) relative to Solarnebula values, 43% and 53% of the oxygen in the chondrulewould come from precursor anhydrous silicates and water

ice in the dust, respectively; only a minor fraction of 4%would be inherited from the Solar gas and organics in thedust (Tenner et al., 2015, see their Table 3). Hence, whenapplying the model originally calibrated for CR chondrulesto chondrules in CV chondrites, new D17O values for theprecursor silicates and water ice need to be defined whereasa difference in D17O values for the organics in both regionscan be ignored. Following Tenner et al., 2015, oxygen iso-tope ratios for both reservoirs are defined based on theassumption that the lowest measured chondrule D17O val-ues (mean of K18 and K23: �8.0‰) are representativefor the effective D17O value of the anhydrous precursor sil-

Fig. 7. Plot showing host D17O values and Mg#’s of chondrules superimposed by oxygen isotope mixing curves of constant dust enrichmentand ice enhancement from the model of Tenner et al. (2015). The model adopts the following D17O values for the various oxygen reservoirs:anhydrous silicate dust, �8.0‰; Solar gas, �28.4‰; water ice, +2.0‰; organics in the dust, +11.3‰. Inferred anhydrous dust enrichment andice enhancement factors for type I chondrules are 50–100� relative to Solar abundance and from anhydrous to �0.6� the water ice content ofdust of CI composition, respectively. Mg#’s and D17O values of the type II chondrules suggest higher: (IIB: �300�, IIA: �2000�) dustenrichment factors and water ice contents (1–4� the amount of water ice of dust of CI composition). The error bars for Mg#’s represent therange (min–max); for host (mean) D17O values, error bars are propagated uncertainties (see Table 1).

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 127

icates and that chondrules with higher D17O values areformed by addition of 16O-poor water ice. Since type IIchondrules likely formed at highly dust-enriched condi-tions, the D17O value of the water ice is calculated usingthe measured D17O value of chondrule K26 (�2.2‰, typeII) in the following way: at a dust enrichment factor of1000, the mass balance involves oxygen from the sourceswater ice, anhydrous silicate dust, and organics in the fol-lowing proportions 54:44:2 (Tenner et al., 2015); solvingthe mass balance for the D17O value of water ice, resultsin a value of +2‰.

Host chondrule D17O and Mg#’s from Kaba are shownin Fig. 7 in combination with modeled dust and ice enrich-ment factors. Error bars for Mg#’s denote maximum-minimum ranges including olivine and pyroxene analysesin individual chondrules of Kaba. Data suggests that higherchondrule Mg#’s correlate with lower D17O values which isin line with results for chondrules from CR chondrites(Schrader et al., 2013, 2014, 2017; Tenner et al., 2015).According to the model of Tenner et al. (2015), values oftype I host chondrules in Kaba correspond to low dustenrichment factors between 50� and 100� and lowamounts of water ice (anhydrous CI dust to 0.6� the nom-inal amount of ice in dust of CI composition) in the dust.The type IIB chondrule could have formed at 300� dustenrichment and �3� water ice. However, formation of thischondrule might be unrelated to that of other chondrulesbecause d18O values are fractionated in such a way thatanalyses plot off the PCM and on the TF line. This distin-guishes them from analyses of all other chondrules and sug-gests that at least one different precursor reservoircontributed to their formation (e.g., Clayton et al., 1983;Tenner et al., 2015). The type II chondrule K26 possibly

formed at dust enrichments of 2000�. Results from thisstudy support the idea that Mg#’s and D17O values maybe correlated even for very Mg-rich chondrules withMg#’s above 99. For these reducing conditions (logfO2:IW – 3.5), increasing D17O values in combination withdecreasing Mg#’s may be due to a small increase in the dustenrichment factor or water ice contents or a combination ofboth parameters (see Fig. 7), as noted by, e.g., Tenner et al.(2015).

Ice enhancement and, to a lower degree, dust enrichmentfactors inferred from chondrule D17O values vary with theassumed isotope ratios of the contributing reservoirs. Forexample, overestimating the D17O value of anhydrous sili-cate dust in the precursor by 5‰ (D17O = �12‰ insteadof �8‰) has only a limited effect on inferred maximumdust enrichments (slight decrease) but increases the maxi-mum estimated ice enhancement factor to 1� the nominalamount of ice in CI dust (type I chondrules), when keepingother parameters constant. As discussed by Tenner et al.(2015), alternative D17O values of water ice (e.g., +80‰)can significantly decrease the inferred ice enhancement fac-tor while only moderately affecting dust enrichment factors,e.g., shifting maximum enrichment factors obtained forKaba type I chondrules from 100� to 200� CI dust relativeto Solar abundance. In conclusions, dust enrichment fac-tors inferred from chondrules in Kaba are likely reasonableestimates for the chondrule forming environment, while iceenhancement factors are less accurately known and sensi-tive to the applied D17O values of reservoirs.

In the scope of the model by Marrocchi and Chaussidon(2015), relict olivine grains represent the direct silicate pre-cursor material for the individual chondrule they areincluded in. Differences between isotope ratios of individual

Fig. 8. Histograms of host D17O values from chondrules in Kabaand further carbonaceous chondrites. Like Kaba, all listedcarbonaceous chondrites contain abundant type I chondrules withD17O values in the range of ��6‰ and �4‰, though no type Ichondrules in Kaba show D17O values >�3‰ that are common inthe other carbonaceous chondrites except Yamato 82094. BecauseAllende experienced significant thermal metamorphism, Mg#’s ofchondrules are disturbed and no distinction is being made betweentype I chondrules with Mg# <98 or >98. Literature data are from[1] Rudraswami et al. (2011), [2] Tenner et al. (2013), [3] Tenneret al. (2015), [4] Tenner et al. (2017), [5] Ushikubo et al. (2012).

128 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

chondrules are, consequently, due to isotopically distinctprecursor olivine rather than due to influence of 16O-poorwater ice. Further, Marrocchi and Chaussidon (2015) sug-gest that, if present, isotopic disequilibrium between chon-drule olivine and pyroxene (D18Opx-ol) represents apossible proxy for dust enrichment in the chondrule-forming region. Accordingly, isotopically homogeneouschondrules would indicate high (>100 times) dust enrich-ments. A D18Opx-ol of approximately +1.5 like observed inchondrules K5 and K13 would amount to dust enrichmentssmaller than 5�, considering precursor olivine of �5‰(Marrocchi and Chaussidon, 2015, see their Table S1).Within the framework of that model, 16O-poor relict olivineas a source for 16O-rich pyroxene as found in chondruleK16 is only conceivable with 16O-rich gas and extremelylow dust enrichment factors (<0.001). The model predic-tions such as the low dust enrichment factors for the threechondrules are based on the explicit assumption that mostolivine are relict grains. Although it cannot be ruled outthat some olivine grains with D17O values similar to pyrox-enes are actually relict grains, there is recent chemical andisotopic evidence from chondrules in CR chondrites(Schrader et al., 2014) that argue against such a scenariowhere most olivine in type I chondrules are relict (e.g.,Libourel and Krot, 2007; Libourel and Chaussidon, 2011).

4.4. Comparison to data from the Allende CV3 and other

carbonaceous chondrites

It can be seen from the previous discussion that hostchondrule D17O values in combination with mean Mg#’sprovide important information about the contribution ofdifferent oxygen reservoirs to chondrule formation. Withhistograms presented in Fig. 8, host chondrule D17O valuesof Kaba are compared to those of various groups of car-bonaceous chondrites in order to identify common chon-drule populations. Literature data in Fig. 8 are selectedbased on the following criterions: (1) The precision of indi-vidual chondrule means are small compared to bin size(<0.5‰), (2) the chondrule means were obtained from mul-tiple analyses (n > 4) of single chondrule at sub‰ precisionin order to exclude obvious relict olivine analysis, and (3)the total number of chondrules in the dataset are statisti-cally meaningful (n � 20) and generally cover a representa-tive set of chondrule types. It is important to note thatfrequencies of host D17O values can be prone to misinter-pretations because available chondrule data may not becompletely representative for the meteorite. In many stud-ies, chondrules are primarily selected based on the size ofmineral grains due to spatial resolution of the SIMSmethod. For example, grain size limitations might havebiased Allende data towards type IA chondrules(Rudraswami et al., 2011). Also, minor chondrule types,such as type II chondrules in carbonaceous chondritesand Al-rich chondrules are often over-selected comparedto their actual representative abundances (e.g., Connollyand Huss, 2010; Ushikubo et al., 2012; Schrader et al.,2013; Tenner et al., 2017).

Histograms of host chondrule D17O values of Kaba,Allende, Yamato 81020, two CRs, Yamato 82094, and

Acfer 094 (for references see Fig. 8) show at least one com-mon mode of D17O values in between �6‰ and �4‰.Yamato 81020, the CR chondrites, and Acfer 094 also con-tain a significant fraction of chondrules with D17O values of��2‰, which are observed in Kaba, Allende, and Yamato82094 as minor components. Except for Yamato 81020, allcarbonaceous chondrites mentioned in Fig. 8 contain chon-drules with D17O values of about 0‰. Only the Kaba, Acfer094, and Yamato 82094 chondrites contain chondrules withD17O values lower than �7‰. In general, type I chondrulesare 16O-rich in comparison to type II chondrules and showa negative correlation between chondrule Mg# and D17Ovalues, as indicated by different grey tones in Fig. 8. More

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 129

precisely, excluding the two CR chondrites, most type Ichondrules possess host D17O values in the range between�6‰ and �4‰ with a minor mode at about �2‰ thatmainly comprises chondrules with Mg#’s below 98. In con-trast to the strong bimodal host D17O values from the Acfer094 and Yamato 81020 chondrites, data on type I chon-drules from the CR chondrites suggest a continuum ofD17O values ranging from �6 to �1‰. At large, it appearsthat type II chondrules only possess D17O values higherthan �4‰ whereas type I chondrules can show host D17Ovalues from �10‰ up to 0‰.

Analyzed chondrules from the Kaba CV3 chondrite dif-fer from those in the aforementioned chondrites, especiallyAllende, in two major aspects: First, host chondrule Mg#’sare consistently above 98 and, second, there exist no type Ichondrules in Kaba that show D17O values higher than��3‰. In Allende chondrules, the difference in Mg#’s isdue to secondary processes, so that high Mg#’s amongmany Kaba chondrules represent the primary charactersof chondrules in CV. However, the lack of type I chon-drules with D17O > �3‰ among chondrules in Kaba stud-ied here could be due to sampling bias because onlychondrules from a single thin section of Kaba were ana-lyzed. Precursors of type I chondrules with 16O-poor signa-ture (D17O > �3‰) are indicated from the analysis ofchondrule K16 (IA, Fig. 1e), in which relatively 16O-richpyroxene analyses define the host value (D17O = �5.03 ±0.27‰, unc.) by excluding 16O-poor (D17O = �1.57 ±0.22‰, unc.) relict olivine analysis. Such 16O-poor relict oli-vine was also found in Allende chondrules (Rudraswamiet al., 2011).

5. CONCLUSIONS

Olivine and pyroxene in chondrules from the KabaUSNM 1052-1 thin section were analyzed for oxygenthree-isotopes (SIMS) and mineral chemistry (EPMA) inorder to evaluate the degree of isotopic homogeneity inchondrules and the relationship of chondrule D17O valuesand Mg#’s. More than one third of the analyzed chon-drules in the Kaba CV3 contain isotopic olivine relict grainsthat are generally 16O-rich relative to chondrule means;however, three chondrule also comprise 16O-poor relictgrains. Excluding isotopic relicts, mean chondrule D17Ovalues were calculated and it was demonstrated for mostof the chondrules that olivine and pyroxene are isotopicallyindistinguishable and, consequently, that both mineralslikely crystallized from the last chondrule-forming melt.

Except for two type II chondrules, the analyzed thin sec-tion only contains Mg-rich chondrules with high and uni-form Mg#’s (host: 99.6–98.8). Consistent Mg#’s ofolivine and pyroxene in individual chondrules support theinterpretation that Kaba was only marginally affected bythermal metamorphism. The majority of type I chondrulespossesses host D17O values in the range between �6‰ and�4‰ and only two type I chondrules show more 16O-richcompositions (��8‰). The type IIB and IIA chondrulesare relatively 16O-poor and yield D17O values of 0‰ and�2‰, respectively. The D17O values of type I chondrules

in Kaba tend to increase continuously with decreasingMg#’s.

In conclusion, type I chondrules in the Kaba CV3 chon-drite are witnesses of an oxygen reservoir with an effectiveD17O value of approximately �5‰ which is also commonlyfound in other groups of carbonaceous chondrites (e.g.,Jones et al., 2004; Krot et al., 2006; Libourel andChaussidon, 2011; Rudraswami et al., 2011; Ushikuboet al., 2012; Tenner et al., 2013, 2015; Schrader et al.,2014). Interestingly, host D17O values of type I chondrulesin Kaba don’t show a pronounced bimodal distributionas it is evident for other carbonaceous chondrites such asAcfer 094 and Yamato 81020. At this point it is not clearwhether the predominance of D17O values in between�6‰ and �4‰ is an exclusive property of the investigatedthin section or representative for Kaba in general.

If the analyzed chondrules in Kaba are representative ofCV chondrites, the correlation of D17O and Mg#’s points tosimilar principals that governed formation of type I chon-drules in CV and CR chondrites. This includes a possibleinvolvement of water ice as an oxidizing 16O-poor agent.Inferred dust enrichment factors for the local disk environ-ment are only moderately high (50–100�) and dust wasalmost anhydrous (anhydrous to 0.6� the nominal ice indust of CI composition), based on investigations of type Ichondrules in the Kaba CV3.

ACKNOWLEDGEMENTS

We thank Glenn MacPherson (National Museum of NaturalHistory, Smithsonian Institution) for the allocation of Kaba thinsection for this study. We are grateful to John Fournelle for helpwith EPMA measurements, Jim Kern for SIMS support, and NoelChaumard for discussion. Reviews of Devin Schrader and twoanonymous reviewers significantly improved the manuscript. Thiswork is supported by NASA Cosmochemistry Program(NNX14AG29G). WiscSIMS is partly supported by NSF(EAR13-55590).

APPENDIX A. SUPPLEMENTARY MATERIAL

Supplementary data associated with this article can befound, in the online version, at https://doi.org/10.1016/j.gca.2017.12.013.

REFERENCES

Alexander C. M. O. ’D. (2004) Chemical equilibrium and kineticconstraints for chondrule and CAI formation conditions.Geochim. Cosmochim. Acta 68, 3943–3969.

Berlin J., Jones R. H. and Brearley A. J. (2011) Fe-Mn systematicsof type IIA chondrules in unequilibrated CO, CR, and ordinarychondrites. Meteorit. Planet. Sci. 46, 513–533.

Bonal L., Quirico E., Bourot-Denise M. and Montagnac G. (2006)Determination of the petrologic type of CV3 chondrites byRaman spectroscopy of included organic matter. Geochim.

Cosmochim. Acta 70, 1849–1863.Busemann H., Alexander C. M. O. ’D. and Nittler L. R. (2007)

Characterization of insoluble organic matter in primitivemeteorites by microRaman spectroscopy. Meteorit. Planet.

Sci. 42, 1387–1416.

130 A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131

Chaussidon M., Libourel G. and Krot A. N. (2008) Oxygenisotopic constraints on the origin of magnesian chondrules andon the gaseous reservoirs in the early Solar System. Geochim.

Cosmochim. Acta 72, 1924–1938.Choi B.-G., Krot A. N. and Wasson J. T. (2000) Oxygen-isotopes

in magnetite and fayalite in CV chondrites Kaba and Mokoia.Meteorit. Planet. Sci. 35, 1239–1248.

Ciesla F. J. (2005) Chondrule-forming processes – an overview. InChondrites and the Protoplanetary Disk (eds. A. N. Krot, E. R.D. Scott and B. Reipurth). Astronomical Society of the Pacific,San Francisco, CA, pp. 811–820.

Clayton R. N. (1993) Oxygen isotopes in meteorites. Annu. Rev.Earth Planet. Sci. 21, 115–149.

Clayton R. N. and Mayeda T. K. (1999) Oxygen isotope studies ofcarbonaceous chondrites. Geochim. Cosmochim. Acta 63, 2089–2104.

Clayton R. N., Grossman L. and Mayeda T. K. (1973) Acomponent of primitive nuclear composition in carbonaceousmeteorites. Science 182, 485–488.

Clayton R. N., Onuma N., Grossman L. and Mayeda T. K. (1977)Distribution of the pre-Solar component in Allende and othercarbonaceous chondrites. Earth Planet. Sc. Lett. 34, 209–224.

Clayton R. N., Onuma N., Ikeda Y., Mayeda T. K., Hutcheon I.D., Olsen E. J. and Molini-Velsko C. (1983) Oxygen isotopiccompositions of chondrules in Allende and ordinary chondrites.In Chondrules and Their Origins (ed. E. A. King). Lunar andPlanetary Institute, Huston, TX, pp. 37–43.

Connolly, Jr., H. C. and Desch S. J. (2004) On the origin of the‘‘kleine Kugelchen” called Chondrules. Chem. Erde - Geochem.

64, 95–125.Connolly, Jr., H. C. and Huss G. R. (2010) Compositional

evolution of the protoplanetary disk. Geochim. Cosmochim.

Acta 74, 2473–2483.Di Rocco T. and Pack A. (2015) Triple oxygen isotope exchange

between chondrule melt and water vapor. Geochim. Cosmochim.

Acta 164, 17–34.Dohmen R. and Chakraborty S. (2007) Fe–Mg diffusion in olivine

II. Phys. Chem. Minerals 34, 409–430.Dohmen R., Ter Heege J. H., Becker H.-W. and Chakraborty S.

(2016) Fe-Mg interdiffusion in orthopyroxene. Am. Mineral.

101, 2210–2221.Donovan J. J. (2015) Probe for EPMA v. 11.1.5.. Probe Software,

Inc., Eugene, OR.Ebel D. S. and Grossman L. (2000) Condensation in dust-enriched

systems. Geochim. Cosmochim. Acta 64, 339–366.Farver J. R. (2010) Oxygen and hydrogen diffusion in minerals.

Rev. Mineral. Geochem. 72, 447–507.Fedkin A. V. and Grossman L. (2006) The fayalite content of

chondritic olivine: obstacle to understanding the condensationof rocky material. In Meteorites and the Early Solar System II

(eds. D. S. Lauretta and H. McSween). University of ArizonaPress, Tucson, pp. 279–294.

Fedkin A. V. and Grossman L. (2016) Effects of dust enrichmenton oxygen fugacity of cosmic gases. Meteorit. Planet. Sci. 51,843–850.

Grossman J. N. and Brearley A. J. (2005) The onset of metamor-phism in ordinary and carbonaceous chondrites. Meteorit.

Planet. Sci. 40, 87–122.Hewins R. H. (1996) Chondrules and the protoplanetary disk: an

overview. In Chondrules and the Protoplanetary Disk (eds. R. H.Hewins, R. H. Jones and E. R. D. Scott). Cambridge Univ.Press, Cambridge, pp. 3–9.

Hewins R. H. and Zanda B. (2012) Chondrules. Meteorit. Planet.

Sci. 47, 1120–1138.

Jones R. H. (1996) FeO-rich, porphyritic pyroxene chondrules inunequilibrated ordinary chondrites. Geochim. Cosmochim. Acta

60, 3115–3138.Jones R. H., Leshin L. A., Guan Y., Sharp Z. D., Durakiewicz T.

and Schilk A. J. (2004) Oxygen isotope heterogeneity inchondrules from the Mokoia CV3 carbonaceous chondrite.Geochim. Cosmochim. Acta 68, 3423–3438.

Kimura M. and Ikeda Y. (1998) Hydrous and anhydrousalterations of chondrules in Kaba and Mokoia CV chondrites.Meteorit. Planet. Sci. 33, 1139–1146.

Kita N. T., Ushikubo T., Fu B. and Valley J. W. (2009) Highprecision SIMS oxygen isotope analysis and the effect of sampletopography. Chem. Geol. 264, 43–57.

Kita N. T., Nagahara H., Tachibana S., Tomomura S., SpicuzzaM. J., Fournelle J. H. and Valley J. W. (2010) High precisionSIMS oxygen three isotope study of chondrules in LL3chondrites. Geochim. Cosmochim. Acta 74, 6610–6635.

Krot A. N. and Nagashima K. (2016) Evidence for oxygen-isotopeexchange in chondrules and refractory inclusions during fluid-rock interaction on the CV chondrite parent body. Meteorit.

Planet. Sci. Suppl. 51, A249.Krot A. N., Scott E. R. D. and Zolensky M. E. (1995)

Mineralogical and chemical modification of components inCV3 chondrites. Meteorit. Planet. Sci. 30, 748–775.

Krot A. N., Petaev M. I., Scott E. R. D., Choi B.-G., Zolensky M.E. and Keil K. (1998) Progressive alteration in CV3 chondrites.Meteorit. Planet. Sci. 33, 1065–1085.

Krot A. N., Fegley B., Lodders K. and Palme H. (2000)Meteoritical and astrophysical constraints on the oxidationstate of the Solar nebula. In Protostars and Planets IV (eds. V.Mannings, A. P. Boss and S. S. Russell). University of ArizonaPress, Tucson, pp. 1019–1054.

Krot A. N., Yurimoto H., McKeegan K. D., Leshin L. A.,Chaussidon M., Libourel G., Yoshitake M., Huss G. R., GuanY. and Zanda B. (2006) Oxygen isotopic compositions ofchondrules. Chem. Erde – Geochem. 66, 249–276.

Krot A. N., Nagashima K., Ciesla F. J., Meyer B. S., Hutcheon I.D., Davis A. M., Huss G. R. and Scott E. R. D. (2010) Oxygenisotopic composition of the sun and mean oxygen isotopiccomposition of the protosolar silicate dust. Astrophys. J. 713,1159–1166.

Kunihiro T., Rubin A. E., McKeegan K. D. and Wasson J. T.(2004) Oxygen-isotopic compositions of relict and host grains inchondrules in the Yamato 81020 CO3.0 chondrite. Geochim.

Cosmochim. Acta 68, 3599–3606.Kunihiro T., Rubin A. E. and Wasson J. T. (2005) Oxygen-isotopic

compositions of low-FeO relicts in high-FeO host chondrules inAcfer 094, a type 3.0 carbonaceous chondrite closely related toCM. Geochim. Cosmochim. Acta 69, 3831–3840.

Libourel G. and Krot A. N. (2007) Evidence for the presence ofplanetesimal material among the precursors of magnesianchondrules of nebular origin. Earth Planet. Sci. Lett. 254, 1–8.

Libourel G. and Chaussidon M. (2011) Oxygen isotopic constraintson the origin of Mg-rich olivines from chondritic meteorites.Earth Planet. Sci. Lett. 301, 9–21.

Libourel G., Krot A. N. and Tissandier L. (2006) Role of gas-meltinteraction during chondrule formation. Earth Planet. Sci. Lett.251, 232–240.

Marrocchi Y. and Chaussidon M. (2015) A systematic for oxygenisotopic variation in meteoritic chondrules. Earth Planet. Sci.

Lett. 430, 308–315.McSween H. (1977) Petrographic variations among carbonaceous

chondrites of the Vigarano type. Geochim. Cosmochim. Acta 41,1777–1790.

A.T. Hertwig et al. /Geochimica et Cosmochimica Acta 224 (2018) 116–131 131

Morris M. A., Boley A. C., Desch S. J. and Athanassiadou T.(2012) Chondrule formation in bow shocks around eccentricplanetary embryos. Astrophys. J. 752, 27.

Nagahara H. (1981) Evidence for secondary origin of chondrules.Nature 292, 135–136.

Nagahara H. and Ozawa K. (2012) The role of exchange reactionsin oxygen isotope fractionation during CAI and chondruleformation. Meteorit. Planet. Sci. 47, 1209–1228.

Nagahara H., Kita N. T., Ozawa K. and Morishita Y. (2008)Condensation of major elements during chondrule formationand its implication to the origin of chondrules. Geochim.

Cosmochim. Acta 72, 1442–1465.Nagashima K., Krot A. N. and Huss G. R. (2015) Oxygen-isotope

compositions of chondrule phenocrysts and matrix grains inKakangari K-grouplet chondrite. Geochim. Cosmochim. Acta

151, 49–67.Nakashima D., Kimura M., Yamada K., Noguchi T., Ushikubo T.

and Kita N. T. (2010) Study of chondrules in CH chondrites – I:oxygen isotope ratios of chondrules. Meteorit. Planet. Sci.

Suppl. 45, A148.Nakashima D., Kita N. T., Ushikubo T., Noguchi T., Nakamura

T. and Valley J. W. (2013) Oxygen three-isotope ratios ofsilicate particles returned from asteroid Itokawa by theHayabusa spacecraft. Earth Planet. Sci. Lett. 379, 127–136.

Rubin A. E. (2000) Petrologic, geochemical and experimentalconstraints on models of chondrule formation. Earth Sci. Rev.

50, 3–27.Rudraswami N. G., Ushikubo T., Nakashima D. and Kita N. T.

(2011) Oxygen isotope systematics of chondrules in the AllendeCV3 chondrite. Geochim. Cosmochim. Acta 75, 7596–7611.

Schrader D. L., Connolly, Jr., H. C., Lauretta D. S., NagashimaK., Huss G. R., Davidson J. and Domanik K. J. (2013) Theformation and alteration of the Renazzo-like carbonaceouschondrites II. Geochim. Cosmochim. Acta 101, 302–327.

Schrader D. L., Nagashima K., Krot A. N., Ogliore R. C. andHellebrand E. (2014) Variations in the O-isotope compositionof gas during the formation of chondrules from the CRchondrites. Geochim. Cosmochim. Acta 132, 50–74.

Schrader D. L., Nagashima K., Krot A. N., Ogliore R. C., Yin Q.-Z., Amelin Y., Stirling C. H. and Kaltenbach A. (2017)

Distribution of 26Al in the CR chondrite chondrule-formingregion of the protoplanetary disk. Geochim. Cosmochim. Acta

201, 275–302.Tenner T. J., Ushikubo T., Kurahashi E., Kita N. T. and Nagahara

H. (2013) Oxygen isotope systematics of chondrule phenocrystsfrom the CO3.0 chondrite Yamato 81020. Geochim. Cos-

mochim. Acta 102, 226–245.Tenner T. J., Nakashima D., Ushikubo T., Kita N. T. and

Weisberg M. K. (2015) Oxygen isotope ratios of FeO-poorchondrules in CR3 chondrites. Geochim. Cosmochim. Acta 148,228–250.

Tenner T. J., Kimura M. and Kita N. T. (2017) Oxygen isotopecharacteristics of chondrules from the Yamato-82094ungrouped carbonaceous chondrite: Further evidence for com-mon O-isotope environments sampled among carbonaceouschondrites. Meteorit. Planet. Sci. 52, 268–294.

Tissandier L., Libourel G. and Robert F. (2002) Gas-meltinteractions and their bearing on chondrule formation. Mete-

orit. Planet. Sci. 37, 1377–1389.Ushikubo T., Kimura M., Kita N. T. and Valley J. W. (2012)

Primordial oxygen isotope reservoirs of the solar nebularecorded in chondrules in Acfer 094 carbonaceous chondrite.Geochim. Cosmochim. Acta 90, 242–264.

Wasson J. T. and Rubin A. E. (2003) Ubiquitous low-FeO relictgrains in type II chondrules and limited overgrowths onphenocrysts following the final melting event. Geochim. Cos-

mochim. Acta 67, 2239–2250.Young E. D. and Russell S. S. (1998) Oxygen reservoirs in the early

solar nebula inferred from an Allende CAI. Science 282, 452–455.

Yurimoto H. and Wasson J. T. (2002) Extremely rapid cooling of acarbonaceous-chondrite chondrule containing very 16O-richolivine and a 26Mg-excess. Geochim. Cosmochim. Acta 66,4355–4363.

Yurimoto H., Krot A. N., Choi B.-G., Aleon J., Kunihiro T. andBrearley A. J. (2008) Oxygen isotopes of chondritic compo-nents. Rev. Mineral. Geochem. 68, 141–186.

Associate editor: Yuri Amelin